Embed Size (px)
Citation preview
The climatic impact of supervolcanic ash blankets
Morgan T. Jones Æ R. Stephen J. Sparks ÆPaul J. Valdes
Received: 8 September 2006 / Accepted: 21 March 2007 / Published online: 22 May 2007
� Springer-Verlag 2007
Abstract Supervolcanoes are large caldera systems that
can expel vast quantities of ash, volcanic gases in a single
eruption, far larger than any recorded in recent history.
These super-eruptions have been suggested as possible
catalysts for long-term climate change and may be
responsible for bottlenecks in human and animal popula-
tions. Here, we consider the previously neglected climatic
effects of a continent-sized ash deposit with a high albedo
and show that a decadal climate forcing is expected. We
use a coupled atmosphere-ocean General Circulation
Model (GCM) to simulate the effect of an ash blanket from
Yellowstone volcano, USA, covering much of North
America. Reflectivity measurements of dry volcanic ash
show albedo values as high as snow, implying that the
effects of an ash blanket would be severe. The modeling
results indicate major disturbances to the climate, particu-
larly to oscillatory patterns such as the El Nino Southern
Oscillation (ENSO). Atmospheric disruptions would con-
tinue for decades after the eruption due to extended ash
blanket longevity. The climatic response to an ash blanket
is not significant enough to instigate a change to stadial
periods at present day boundary conditions, though this is
one of several impacts associated with a super-eruption
which may induce long-term climatic change.
Keywords Supervolcano � Super-eruption � Ash blanket �ENSO � Climate change
1 Introduction
Supervolcanoes are caldera systems capable of erupting in
excess of hundreds to thousands of km3 of magma in a
single event. Examples include Yellowstone in USA, La
Pacana in Chile and Toba in Indonesia. Large eruptions are
classified on their total ejected mass (m) using a logarith-
mic magnitude scale, M (Pyle 1995; Pyle 2000). This is
defined by: M = log10 (m) – 7.0. Explosive super-eruptions
are defined as M8 or larger by Mason et al. (2004).
Eruptions of this magnitude are rare, with a minimum
estimated frequency of 1.4 events Ma–1, though multiple
super-eruptions appear to occur in pulses of activity (Ma-
son et al. 2004). A recent example of an eruption of this
magnitude came from the Toba caldera in Sumatra around
74,000 years ago (Rose and Chesner 1987; Rampino and
Self 1993; Zielinski et al. 1996; Oppenheimer 2002). Su-
pervolcanoes are predicted to have large magma supply
rates up to and possibly above 107 m3 s–1 (Baines and
Sparks 2005). Rapid injection of material into the atmo-
sphere entrains and heats the surrounding air. The resulting
effects of expansion and dilution significantly reduce the
density of the eruption cloud and give the jet buoyancy.
The cloud does not become buoyant immediately as the
width of the vent prevents effective mixing of the atmo-
sphere with the centre of the eruption column. This leads to
column collapse and the formation of radial pyroclastic
density currents. The deposits from these events are known
as ignimbrites. Entrainment of air at the top of the flow and
sedimentation of larger lithic (rock) fragments reduces the
net density as the flow progresses. This allows the upper
M. T. Jones (&) � R. S. J. Sparks
Department of Earth Sciences, University of Bristol,
Wills Memorial Building, Queens Road,
BS8 1RJ Bristol, UK
e-mail: [email protected]
P. J. Valdes
School of Geographical Sciences, University of Bristol,
University Road, BS8 1SS Bristol, UK
123
Clim Dyn (2007) 29:553–564
DOI 10.1007/s00382-007-0248-7
part of the flow to become buoyant and rise (Sparks and
Walker 1977; Sparks et al. 1986). There may be fluctua-
tions in magma supply rates during ignimbrite formation,
but observations from previous eruptions, theoretical work
and experimental analysis suggest that the lift-off of a
co-ignimbrite cloud is relatively abrupt and can therefore
be assumed to act as a single cloud (Sparks et al. 1986;
Sparks and Bonnecaze 1993; Sparks et al. 1997). The
supply rates of ash and hot air into the rising plume are
predicted to be 1011–1013 m3 s–1 (Baines and Sparks
2005). The ash forms an umbrella cloud when it reaches
neutral buoyancy, predicted to be between 27 and 38 km
above sea level at the peak of the eruption (Woods and
Wohletz 1991; Rampino and Self 1992, 1993; Baines and
Sparks 2005).
If ash umbrella clouds become over 600 km in diameter
the rate of spreading becomes controlled by a balance be-
tween gravity and Coriolis forces (Baines and Sparks
2005). The ash cloud forms a spinning body that has nearly
fixed proportions and is largely unaffected by stratospheric
winds, allowing the cloud to spread out radially. The re-
moval of ash by gravitational sedimentation in the days and
weeks after the eruption creates a radial ash blanket around
the vent. A typical super-eruption can expel around
1,000 km3 of co-ignimbrite ash, enough to cover a conti-
nent the size of North America (~107 km2) with a blanket
of ash 100 mm thick.
2 Methodology and approach
Several processes in super-eruptions can severely disturb
the ocean-atmosphere system. The most widely studied is
the mass injection of ash, SO2 and aerosols into the
stratosphere. For example, the 74 ka Toba eruption is
estimated to have ejected 1012–1013 kg of volcanic aero-
sols and ash (Rampino and Self 1992). The effects of
aerosols in the stratosphere have been highlighted as the
primary mechanism for climate forcing on timescales of a
few years (Rampino and Self 1992, 1993; Bekki et al.
1996; Zielinski et al. 1996; Carroll 1997; Robock 2000;
Rampino 2002; Jones et al. 2005; Timmreck and Graf
2005). In particular, sulfuric acid aerosol scatters short-
wave radiation and causes a net cooling of the Earth’s
surface (Bekki 1995; Robock 2000; Savarino et al. 2003;
Jones et al. 2005; Timmreck and Graf 2005). Ash has a
stratospheric residence time of a few weeks and long-term
climatic effects from ash in the atmosphere are believed to
be negligible. SO2 has a slightly longer residence time as
its removal is dependent on conversion to H2SO4 aerosol.
Excessive SO2 loading can potentially reduce OH levels,
leading to stratospheric dehydration and an increased SO2
lifetime by impeded H2SO4 conversion (Bekki 1995).
However, stratospheric aerosol concentrations are lower as
a result and the entrainment of tropospheric water has not
been included in previous models. Timescales for the
influence of a sulfate cloud depend on the post-eruptive
hydration state of the stratosphere. Unrestricted atmo-
spheric residence times for SO2 are between 1 and 3 years
(Robock 2000). If complete dehydration of the stratosphere
should occur, the climate may be directly affected for
several years longer (Bekki 1995; Bekki et al. 1996).
Volcanogenic processes capable of altering the climate
on longer timescales include the deposition of ash on land.
Vegetation proximal to the volcano will be buried by ash.
10 mm of ash is enough to disrupt most forms of agricul-
ture and flora (Sparks et al. 2005). Plants and trees too high
to be buried are often killed by the effects of acid rain and
ash-leachates. Volcanic gases, metal salts and other vola-
tiles erupted can be scavenged from the atmosphere by fine
ash particles. These ash-leachates stay adsorbed onto the
ash particle surfaces until removed by dissolution (Frogner
et al. 2001; Witham et al. 2005). Temperatures in the
stratosphere are too low to support liquid H2O, so most of
the ash-leachates will still be present on the particle surface
when deposited, allowing them to be released into the
surrounding environment. Toxic levels of aluminum (as
AlFx+3–x complexes), fluoride and other compounds
released by the ash induce ecosystem stress and species
decline. This particularly affects systems with a low turn-
over rate, such as ponds, lakes and soils (Frogner 2004).
Ecological sensitivity to ash-leachates depends on the
ability of the ecosystem to buffer the effects. Though
parent magmas of supervolcanoes are volatile-poor com-
pared to basic magmas, a large-scale loading of ash per unit
area will induce toxic and acidic conditions above most
ecosystems’ buffering capacity, killing flora that survived
the burial phase. Such large-scale deforestation will inter-
play with the radiation budget, the carbon cycle and
evapotranspiration.
Ecological responses to the 1883 eruption of Krakatoa
(Indonesia) show that it took 5 years before grasses began
re-colonizing affected islands and 15 years to re-establish
trees and shrubs (Bush 2006). Evidence from the re-colo-
nization of flora after the 1980 eruption of Mount St. He-
lens (USA) shows that pioneer species can become
restricted or even halted by insect herbivory (Fagan and
Bishop 2000; Bishop 2002; Knight and Chase 2005), fur-
ther lengthening the recovery of flora after an eruption. To
our knowledge there are no studies of the decline in flora
following a super-eruption, but the response is likely to last
for much longer as re-colonization of pioneer species re-
quires seeding by either wind or animal transport. The
scale of the deposits dictates that the recovery of flora must
554 M. T. Jones et al.: The climatic impact of supervolcanic ash blankets
123
do so from the edges and heavily eroded areas of the ash
blanket. Climatic responses to vegetation loss may also
hamper the recovery of flora. Therefore, it is likely that an
area in the region of 3 · 106 km2 would be completely
devoid of plant-life following a super-eruption for well
over a decade.
A continental sized ash blanket will interplay with the
surface radiation budget and therefore interact with atmo-
spheric circulation. Volcanic ash is extremely reflective
due to large concentrations of silica rich glass and high
particle vesicularity. As a result, an ash blanket will sig-
nificantly increase the surface albedo. This will reflect a
greater proportion of incoming radiation, leading to surface
cooling. Some tephra deposits have albedo values compa-
rable to snow. Thus the albedo effects of a continental ash
blanket are comparable to those of an ice sheet. Surface
cooling could also initiate the build up of snow and ice on
top of the ash blanket, amplifying the initial albedo forcing.
Residence times of ash blankets at the surface are aided
by the rapid establishment of stable channel and rill (riv-
ulet) networks, decreasing the erodibility of the tephra
layer (Collins and Dunne 1986; Leavesley et al. 1989).
Co-ignimbrite ash is cohesive, inhibiting transportation by
water and increasing surface runoff (Nammah et al. 1986).
A mortar like crust is formed by desiccation, restricting
erosion after the first couple of years to the channel and rill
sides. For example, erosion of tephra from the 1980
eruption of Mount St. Helens slowed from 26.0 to 1.8 mm/
year within 3 years (Collins and Dunne 1986). The scale of
supervolcanic ash deposits should increase the residence
time by at least an order of magnitude. Ash blanket life-
times should be further aided by changes in the hydro-
logical cycle. Surface cooling effects associated with the
aerosol forcing reduce both evaporation rates and the sat-
uration mixing ratio of water vapor in air. Modeling of
such effects at a high sulfur loading show a halving of
average initial global rainfall after an eruption (Jones et al.
2005). A reduction in precipitation will in turn reduce the
erosion rate of the ash blanket. Studies of sedimentary and
geomorphic responses to ignimbrite sheet emplacement
suggest that a large ash blanket could exert a significant
environmental influence for considerably more than
10 years following an eruption (Manville 2002). A possible
negative feedback is the impact of larger volumes of snow
from aerosol induced cooling, which would melt during
spring and increase erosion. Though the lifetime of an ash
blanket is difficult to constrain and the removal of ash at
the surface would be gradual, we surmise that an ash
blanket will remain a significant climatic forcing for at
least 10–50 years after an eruption. The impact of an ash
blanket is likely to alter the climatic responses predicted by
models only dealing with the effect of aerosols and is
therefore an important consideration.
3 Climate model and experimental design
To assess the impact of a known supervolcano, we con-
ducted a climate sensitivity experiment by simulating the
effects of an ash blanket from an eruption of the Yellow-
stone Caldera in Wyoming, USA, using the Hadley Centre
coupled atmosphere-ocean model, HadCM3 (Gordon et al.
2000; Pope et al. 2000). This is a well-known General
Circulation Model (GCM) that has been used to success-
fully predict past and present climatic conditions (Stott
et al. 2000; Hewitt et al. 2001; Tett et al. 2002) and is
widely used for future predictions (e.g. Johns et al. 2003).
Our control simulation is for pre-industrial conditions
(CO2 = 290 ppmv, solar irradiance = 1,365 W m–2). We
use the Meteorological Office Surface Exchange Scheme
(MOSESII), which represents the transfer of heat,
momentum, moisture and carbon between land and atmo-
sphere. The atmosphere has a horizontal resolution of 2.5�by 3.75� and has 19 vertical levels. The ocean has a hori-
zontal resolution of 1.25� by 1.25� and 20 vertical levels.
We have based the modeled ash blanket on ashfall from the
2 and 0.64 Ma eruptions of Yellowstone caldera (Fig. 1)
(e.g. Reynolds 1975; Perkins and Nash 2002; Sparks et al.
2005). This involved two alterations to the model set-up;
the vegetation coverage and the soil albedo. Based on the
inferences from smaller eruptions, we have assumed
complete destruction of vegetation for an area of
3 · 106 km2, labeled the Inner Zone in Fig. 1. In the larger
Outer Zone (~7.56 · 106 km2) 50% of flora is removed,
Fig. 1 Map of North America illustrating the known ashfall from two
eruptions of Yellowstone and the affected area in the ash blanket
simulation. The shaded areas show the ‘two box’ configuration
assumed by the model. For the inner box, total destruction of
vegetation is assumed. The soil values in the model are replaced with
the physical properties of ash. The distal outer box suffers a 50%
reduction in flora. Soil properties for the outer box are an average
between ash and the original soil. The two dashed lines enclose sites
where previous ash deposits have been recognized (e.g. Sparks et al.
2005). The actual area covered by ash is likely to have been
significantly larger
M. T. Jones et al.: The climatic impact of supervolcanic ash blankets 555
123
while maintaining the ratios between original vegetation
types.
The bare soil albedo is the model representative of the
reflectivity of the ground where there is no water or vege-
tation coverage. To ascertain a suitable input parameter,
measurements of ash albedo were made using dry and wet
samples from the Bishop Tuff, USA and the Sifon Ignim-
brite, Chile (Fig. 2), two deposits from previous super-
volcanic eruptions. The two ash samples can be considered
end members in the albedo range for supervolcanic ash,
with the Bishop Tuff more comparable to tephra from
Yellowstone as it is a co-ignimbrite ash sample of similar
composition. The Sifon Ignimbrite is a deposit from
pyroclastic density currents which have a greater compo-
nent of lithic and crystalline material, reducing the albedo.
Reflectivity between 350 and 1,200 nm is averaged by
spectroscopy to give a mean albedo as the model requires a
single number as an input parameter. Dry Bishop Tuff ash
has the highest albedo value of 0.74, equivalent to fresh
snow and much higher than typical values for soils and
sand (0.1–0.35). Figure 2 illustrates the dependency of al-
bedo on the degree of wetness, suggesting a complicated
feedback mechanism between surface albedo and the
hydrological cycle. Soil albedo is not a function of soil
moisture in HadCM3 so we chose an albedo value of 0.47,
the mean of all samples measured and the suggested
method for modeling all soils except for semi-arid and
desert environments (Wilson and Henderson-Sellers 1985).
While the effects of an ash blanket may induce semi-arid
conditions, taking the average of dry and wet samples will
at worst underestimate the mean surface albedo. Additional
experiments using an Atmospheric General Circulation
Model (AGCM) with a slab ocean were run to test the
sensitivity of the model to changes in albedo (see Model
Sensitivity, Chap. 4.4).
The aim of this study is to isolate the potential impacts
of a supervolcanic ash blanket and assess the climatic
sensitivity to a terrestrial forcing unrelated to the effects of
stratospheric aerosols. It is therefore a sensitivity experi-
ment and is not aimed at reproducing a specific super-
eruption. The simulations do not include the effects of
stratospheric loading of aerosols on the climate, allowing
the effects of the ash blanket to be clearly recognized.
Excluding the effects of initial stratospheric loadings
means the starting conditions of the simulations do not
reflect the true initial conditions after a super-eruption,
such as increased snow cover and a reduction in radiative
forcing. Further work will need to couple the effects to-
gether to investigate non-linear feedbacks. For these
experiments, a transient analysis of the climatic response
seems inappropriate. Instead, it is assumed that the lifetime
of the ash blanket is of sufficient length to outlast the
aerosol induced effects and to allow a temporary climatic
steady-state to develop. This model sets out to predict what
these steady-state conditions will be following the relaxa-
tion of the sulfur loading. The control and altered simula-
tions are started from the spun-up control run. In the altered
simulation affected by ash, emplacement of the blanket is
assumed to be instantaneous. The model then simulates
200 years of ashfall affected atmospheric and ocean cir-
culation. The first 20 years are discounted as the spin-up
phase, where the system is trying to attain a new equilib-
rium. The following 180 years are then averaged to give
mean altered climatic conditions. The same procedure is
followed for the control run, allowing comparisons to be
made between the two. This does not assume that the ash
blanket residence time is 200 years, nor that vegetation
growth will remain static during this period. The length of
the simulation is purely to increase the signal to noise ratio
(SNR) in response to the forcing. The key results from our
simulation establish themselves relatively quickly and are
noticeable within a decade. Plots of model responses show
values at the 95% confidence level, based on the student’s t
test. This method approximates regions of statistical con-
fidence but is limited due to issues of temporal and spatial
autocorrelations. All of the major features in the results are
identified at the 99% confidence level.
4 Results
4.1 Circulation and temperature response
The presence of an ash blanket induces large deviations
from control conditions. Surface temperatures show sig-
nificant local variations of ±5�C with only a 0.1�C decrease
on a global scale compared to the control run. The local-
ized character of the response is also true of precipitation,
Fig. 2 Albedo ranges of ash. Dry and wet samples from the Sifon
Ignimbrite (Chile) and the Bishop Tuff (USA) were analyzed by
spectroscopy. For the wet samples, water was added till the ash
reached saturation then a paper towel was dabbed to remove excess
surface water. This gives end member values for degree of wetness
for ash samples
556 M. T. Jones et al.: The climatic impact of supervolcanic ash blankets
123
evaporation and cloud cover. The globally averaged change
in radiative forcing due to the imposed ash albedo was
calculated using the method of Gregory et al. (2004). This
showed that the instantaneous change was 1.0 W m–2 and
that the model regained energy balance within the
200 years of the simulation. Locally over the ash blanket,
the initial change in net top-of-the-atmosphere outgoing
solar radiation is approximately 32 W m–2. The combina-
tion of deforestation and an increase in albedo instigates
local surface cooling exceeding 5�C throughout the year in
North America, well outside the natural variability of the
control model (Fig. 3). Surface cooling leads to higher
surface pressures, especially during June, July and August
(JJA), as increased insolation strengthens the absolute
albedo effect (Fig. 4). The local decrease in temperature in
North America changes the temperature gradient with the
adjacent North Pacific. The heat stored by the ocean warms
the atmosphere during December, January and February
(DJF), causing an increase in baroclinic instability and the
formation of more frequent and/or deeper low-pressure
systems over the North Pacific (based on band passed fil-
tered eddy kinetic energy, not shown). A greater response
is observed during JJA due to the peak in relative surface
cooling in North America at maximum insolation. Changes
in pressure interplay with the northern hemisphere jet
streams. The simulation results indicate a latitudinal shift
of storm tracks in both the North Pacific and the North
Atlantic Oceans (based on band pass filtered eddy kinetic
energy). Comparisons of the perturbation to the stream
function at 200 hPa between the altered simulation and the
control suggest southwards movement in the North Pacific
and northwards movement in the North Atlantic. In the
North Pacific the storm track is amplified by up to 5 m2 s–1,
presumably due to less frequent blocking by high-pressure
systems. As a consequence slightly warmer conditions are
promoted in northern North America compared with the
control (Fig. 3).
The Pacific/North American teleconnection pattern
(PNA) is affected by the surface perturbation in North
America. The PNA is closely linked to the El Nino
Southern Oscillation (ENSO) and variability in ENSO re-
gion 3.4 shows a consistent increase in inter-annual vari-
ability of sea surface temperatures (Fig. 5). HadCM3 is
considered relatively accurate at mimicking natural vari-
Fig. 3 Comparisons of surface
air temperature between the ash-
affected simulation and the
control run (�C), showing
anomalies exceeding the 95%
statistical confidence level using
the student’s t test. a Show the
global temperature changes for
JJA and b highlights the
temperature difference during
DJF
M. T. Jones et al.: The climatic impact of supervolcanic ash blankets 557
123
ability in ENSO (Collins 2000; Collins et al. 2001). Vari-
ability rises by more than 25% in all months, reaching a
maximum of a 60% increase in December. The dominant
change in ENSO variability is the magnitude of El Nino
and La Nina events, with mean monthly sea surface tem-
peratures ranging from 23 to 29�C in ENSO region 3.4
(Fig. 6). Enhancement of ENSO variability is not matched
by large changes in eastern Equatorial Pacific surface
temperatures (Fig. 3), so increases in frequency and/or
magnitude of El Nino events are offset by La Nina epi-
sodes. The Southern Oscillation Index (SOI) provides an-
other indicator of changes to ENSO by comparing the
mean pressure changes for Tahiti and Darwin (Australia),
with negative values suggesting a bias towards El Nino
phases. Mean pressures drop in Tahiti from 1,009.9 to
1,009.5 mbar (2r = 1.0) and rise in Darwin from 1,003.6
to 1,006.7 mbar (2r = 1.3), combining to give a DSOI of
–3.5. This suggests the ENSO signal does show greater or
more frequent El Nino phases. Model predictions of ENSO
should be treated with some caution due to large decadal
and centurial variations in variability and differing
responses of climate models to climate forcings (Collins
2000). However, the large magnitude of the response to an
ash blanket gives us confidence in the validity of our
results.
The changes in the atmospheric circulation have sig-
nificant repercussions throughout the globe. Northern
Fig. 4 Comparisons of mean
sea level pressure between the
altered run and the control run
(mbars), showing anomalies
exceeding the 95% statistical
confidence level using the
student’s t test. a Shows
conditions during JJA and bshows the change during DJF.
High latitude pressures increase
during summer months and
decrease during winter months
in both hemispheres
Fig. 5 Comparisons of surface temperature variability in ENSO
region 3.4 between the control run and the altered simulation. The
minimum rise in variability is in August (25%) and maximum
variability occurs in December (60%)
558 M. T. Jones et al.: The climatic impact of supervolcanic ash blankets
123
hemisphere continents unaffected by ash burial are warmed
by the increase in the advective effects transporting heat
from the oceans to the continents. The warming is slight, so
only certain areas such as northwest Russia and northeast
USA see temperature increases outside the natural vari-
ability of the system (Fig. 3). However, there is sustained
cooling of northern hemisphere continents during JJA that
is greater than the 95% statistical confidence range. The
Middle-East, northeast Africa and Eurasia all display 1�C
cooling in response to changes in atmospheric circulation.
There are significant changes in surface temperature and
mean sea level pressure in the southern hemisphere, but
these are restricted to oceans.
4.2 Ocean response
Ocean surface temperatures also react to the ash blanket
forcing, though to a lesser degree than the atmospheric
response. The Norwegian Sea shows a consistent and
sustained cooling, where temperatures are consistently 1�C
colder than the control. This should lead to large and
sustained increases in sea ice. Svalbard (Norway) experi-
ences a 42% increase in mean sea ice cover to 76.8% from
the control value of 34.9% during DJF (2r = 30.8). Further
south, temperatures are influenced by the strengthening of
the North Atlantic Thermohaline Circulation (NATHC).
Overturning circulation shows a uniform increase from
16.8 to 19 Sv throughout the 200 year simulation, raising
ocean temperatures by 2�C around Newfoundland
throughout the year and by 1�C around the British Isles
during JJA. Winter ocean temperatures in the equatorial
Pacific are raised by 1–2�C in response to the increase in
ENSO variability, supporting the suggestion of a slight bias
towards El Nino episodes in ENSO. There is a comparable
decrease in eastern North Pacific temperatures in response
to the rearrangement of atmospheric circulation. Elsewhere
ocean surface temperatures are relatively unaffected and
show no long-term trends through the length of the simu-
lation.
4.3 Precipitation response
Global precipitation patterns are modified by the changes
in atmospheric circulation. In North America precipitation
is reduced by over 2 mm/day during JJA from an average
of 2.68 mm/day (Fig. 7). Evapotranspiration accounts for
around 21% of precipitation recycled in the Mississippi
Basin (Trenberth 1999), so the loss of vegetation accounts
for some of the decrease. A reduction in the latent heat flux
associated with deforestation slows the hydrological cycle.
Other factors include the increase in surface pressures
stabilizing the atmospheric boundary layer, possibly by
reducing the air mass internal convection, which aids the
reduction in precipitation. Precipitation in North America
is less affected during DJF, which significantly increases
snowfall due to the sustained drop in temperature. Recur-
rent snowfall covers large expanses of the ash blanket from
November to April. The North American mean snow cover
is greater than 50 cm thick above 40�N and the surface
albedo suggests that the entire ash blanket is covered
during February. The enhanced snow cover provides a
positive feedback to the surface albedo increase and the
subsequent impact on atmospheric circulation.
Around the Tropic of Cancer there is a general decrease
in precipitation on continents, with Central America, the
Sahel and Southeast Asia all seeing rainfall drop by around
1 mm/day during JJA (Fig. 7). The reduction in precipi-
tation around the Bay of Bengal is probably attributed to
the influence of ENSO on the Indian Monsoon, though the
teleconnection is not a stable one. The strength of the
monsoon is often diminished during El Nino years, so the
reduction in precipitation in Southeast Asia may well be
linked to the slight preference to El Nino phases in ENSO.
The changes in the Americas and Africa are more difficult
to explain, but are most likely to be in response to summer
cooling of mid- to high-latitude continents. Monsoonal
weather patterns gain strength from the temperature gra-
dient between land and ocean, so a summer cooling of
continents serves to reduce the strength of tropical weather
systems. The precipitation decreases exceed the maximum
variance of the system over much of Africa and India,
suggesting an increase in drought frequency (Fig. 7a).
During DJF the precipitation changes over the continents
are much smaller (Fig. 7b). The main reason for the dis-
Model Years
220 20 40 60 80
24
26
28
30S
urfa
ce T
empe
ratu
re (
°C)
Control SimulationAsh Layer Simulation
Fig. 6 Surface air temperatures for all months of the run in ENSO
region 3.4, with the mean seasonal cycle removed. The increase in
ENSO variability is dominated by an increase in El Nino and La Nina
magnitudes. Extended periods of ‘normal’ circulation conditions are
much less frequent in the ash layer simulation
M. T. Jones et al.: The climatic impact of supervolcanic ash blankets 559
123
crepancy between the seasons is the positioning of the
continents. Maximum insolation during DJF is in the
southern hemisphere, where there are less continental land
masses. Changes in seasonal temperature fluctuations af-
fect continents more than oceans, so the impact on conti-
nental precipitation patterns during DJF is not as severe. In
contrast, ocean precipitation patterns are affected
throughout the year. Increases of over 2 mm/day 10� north
and south of the equator are observed all year in the Pacific
in response to increases in ENSO variability and changes to
either the position and/or variability in the position of the
Inter-Tropical Convergence Zone (ITCZ).
4.4 Model sensitivity
We conducted a series of simplified model simulations in
order to assess the sensitivity of the results to vegetation
loss and albedo increase. Due to computing time con-
straints the test runs used a slab ocean model (HadSM3)
with the same atmospheric model component in HadCM3.
These simulations simplify the calculations by assuming a
thermodynamic ocean with predefined circulation and heat
transport. Atmospheric responses to an ash blanket in
HadSM3 compare well to those derived from the HadCM3
simulation. The only major exception is the response of the
equatorial Pacific, where the ash blanket forcing manifests
a response similar to ENSO by raising the surface air
temperatures by 5�C in the slab ocean models as opposed
to increasing ENSO variability (Fig. 8). The slab model is
incapable of reproducing changes in oceanic overturning,
but the increase in equatorial surface air temperatures
suggests that it picks up the atmospheric forcing that
interacts with ENSO in the coupled model. While the style
of the response is comparable to the coupled model, the
slab model appears to exaggerate the magnitudes of
changes. Three experiments were conducted at varying
albedo values to compare with a control simulation. Run ahad an ash blanket albedo of 0.47, run b slightly lower at
0.36 (the average value for Sifon ash, Fig. 2) and run cused the original soil input parameters to assess the impact
of vegetation loss alone.
The results suggest that both albedo increase and
vegetation loss are important climate forcings. DJF tem-
peratures in North America drop by over 6�C in all three
simulations, regardless of surface albedo (Fig. 8). Vegeta-
tion removal exposes bare earth and reduces surface heat
Fig. 7 Comparisons of mean
precipitation between the ash-
affected simulation and the
control run (mm/day), showing
anomalies exceeding the 95%
statistical confidence level using
the F test. a Show the global
precipitation changes for JJA
and b shows the precipitation
difference for DJF
560 M. T. Jones et al.: The climatic impact of supervolcanic ash blankets
123
retention. In response, the surface albedo during the winter
increases through sustained snow accumulation above that
of the control. This mechanism seems to be the dominant
forcing, as the climatic response in run c during DJF is very
similar to the other two models. In contrast, the model
responses during JJA are sensitive to surface albedo. In run
c, the planetary albedo over North America during JJA is
lower than the control through exposure of dark soils and a
reduction in cloud cover over North America. Changes in
albedo are insufficient to explain adequately the rise in
North American surface temperatures in run c however,
which peak at 9�C above the control (Fig. 8). The main
reason for different responses is attributed to the interaction
of evapotranspiration losses and surface albedo. Vegetation
loss effectively reduces atmospheric moisture supply. The
mean latent heat flux in central North America decreases
from 140–160 W m–2 in the HadCM3 control to 20–
40 W m–2 in the HadCM3 ash-affected simulation, also
observed in the three simplified simulations. The large
decrease in latent heat flux leads to surface heating during
maximum insolation through elevated convective heating,
as seen in run c. The prevalence of a net surface cooling in
run a and the main coupled simulation during JJA suggests
that the surface cooling from the increase in surface albedo
overprints the surface heating from the reduction in latent
heat flux. Run b appears to be a combination of the
processes, with net surface heating above 35 N and net
cooling further south. This suggests that a high surface
albedo becomes dominant in areas or times of greater
insolation. Overall the similarities in DJF weather condi-
tions between runs a, b and c suggest that the effects of
vegetation loss dominate in winter conditions (Fig. 8). In-
creases in surface albedo only dominate the local response
during maximum insolation (JJA).
Overall the sensitivity experiments imply that both in-
creased surface albedo and vegetation loss instigate a large
climatic response. Varying the surface albedo does not
result in large annual global changes in temperature, but
does affect temperature variability in the vicinity of the ash
blanket. A high surface albedo has the greatest impact
during periods of high insolation and at low latitudes. The
net impact is local cooling, initiating a global response.
Combined, the effects of vegetation loss and albedo in-
crease have a large impact on circulation patterns without
dramatically changing the global temperature. These sen-
sitivity studies highlight the difficulty in accurately
assessing the transient response to a super-eruption. In the
first few years after a super-eruption one would expect the
influence of vegetation loss to dominate over the albedo
increase due to reduced radiative forcing, but both pro-
cesses are likely to be minor compared to the effect of
stratospheric aerosols. If the erosion of the ash blanket
Fig. 8 Comparisons of mean
surface air temperature change
between HadSM3 simulations.
Row 1 shows run a, with an ash
blanket albedo of 0.47. Row 2shows run b with an albedo of
0.36. The 3rd row shows the
results of run c, which only
simulates the removal of
vegetation and keeps the
original model surface albedo
values. The left column is the
average of DJF, the rightcolumn the mean of JJA months
in the runs
M. T. Jones et al.: The climatic impact of supervolcanic ash blankets 561
123
exceeds the rate of vegetation recovery, one would expect a
transient change in climatic forcing from conditions pre-
dicted by run a to those calculated by run c.
5 Discussion and conclusions
This study has investigated the volcanogenic forcing due to
ash covering a continent sized area. The residence time of
the ash blanket is expected to be decades and thus capable
of perturbing climate well beyond the few first few years,
when the effects of aerosols are likely to be dominant. A
simulated ash blanket from Yellowstone has a significant
impact on atmospheric circulation. The global climate does
not change significantly in the ash-affected simulation, but
seasonal climatic variability is substantially increased. Our
findings, such as the change in ENSO variability and
magnitude, are significant enough to impact climatic re-
sponses to other volcanogenic forcings.
Our choice of supervolcano location and initial starting
conditions may affect climatic response. Similarly, the
response may be somewhat different if the ‘‘background’’
climate was equivalent to past eruptions from Yellowstone
(e.g. 0.64 or 2 Ma), but we suggest that our results showing
the climatic importance of an ash blanket will be robust
across all time periods. Ash blankets from eruptions at low
latitudes will favor albedo induced surface cooling, while
mid-to high latitude eruptions will be affected by seasonal
responses with changes in latent heat flux associated with
vegetation loss playing a more dominant role. Yellowstone
is situated close the Pacific. Thus, the marked ENSO re-
sponse to an ash blanket may be unique to North American
supervolcanoes. The starting conditions are another caveat
to the model response. The decision to use a pre-industrial
control makes this sensitivity study applicable to the cli-
matic response during a warm interglacial period. Initial
ocean circulation is also an important starting parameter,
especially given the response of the NATHC.
The climatic response to a continental ash blanket is less
severe than those predicted for stratospheric sulfur loading.
The response to a sulfate cloud from a Toba sized eruption
is predicted to be global cooling between 3 and 10�C
(Rampino and Self 1992, 1993; Rampino 2002; Jones et al.
2005). The climate sensitivity to an ash blanket does not
include lasting effects from feedbacks other than those
investigated here. There are several volcanogenic processes
that could destabilize the climate prior to the appearance of
effects caused by the ash blanket. Massive cooling events
can dramatically affect ecosystems and flora. An increase
in hard freezes will destroy significant amounts of vege-
tation (Rampino and Ambrose 2000). Further loss of biota
will instigate a large climatic response as the model ap-
pears particularly sensitive to changes in vegetation cover.
Changes in atmosphere chemistry following a super-erup-
tion will also affect the response. Any changes in strato-
spheric chemistry would impact on sulfate concentrations
and lifetimes, and may also affect changes in cloud cover.
The aerosol induced temperature anomaly is predicted to
be 2�C below normal 10 years after the eruption and 0.3�C
cooler after 50 years from (Jones et al. 2005). If these
predictions are correct, the steady-state climate predicted
by our model will be slightly cooler, though this will aid
the ash blanket residence time by slowing the hydrological
cycle.
The impact of cooling on the hydrological cycle has
further implications for the climatic response than just ash
blanket longevity. The NATHC almost doubles in intensity
in response to increased salinity and amplification of
cooling at high latitudes (Hewitt et al. 2001; Luder et al.
2003; Jones et al. 2005). Our results have shown a similar
NATHC response on a smaller scale, which would act as a
positive feedback to the predicted aerosol induced increase
in overturning. The promotion of sea ice formation in the
Labrador and Norwegian Seas is likely to accentuate the
response of the NATHC, though the expansion of sea ice
will also decrease high latitude evaporation. The pertur-
bation of the NATHC is predicted to last for 20 years from
an aerosol forcing (Jones et al. 2005), but the effects of an
ash blanket may heighten the intensity and longevity of the
response, maintaining the NATHC as a negative feedback
to the global cooling and buffering temperature decreases
in western Europe.
Supervolcanoes have been implicated as possible cata-
lysts for longer term climate change during periods of
climatic instability (Rampino and Self 1992, 1993; Am-
brose 1998; Rampino and Ambrose 2000), though this is
contentious (e.g. Oppenheimer 2002; Lee et al. 2004). It
was suggested that global cooling significantly increases
snow and ice cover, raising the surface albedo and pro-
viding a positive feedback to the volcanic cooling. The
response to an aerosol forcing predicts that the decrease in
temperatures are insufficient to initiate a glaciation inde-
pendently (Jones et al. 2005). A prolonged ash blanket
raises the surface albedo considerably and increases snow
and ice cover over North America in winter. It is possible
that the combination of the ash blanket and the aerosols
may allow persistent snow cover and encourage the
expansion of ice sheets, but the study of their isolated
impacts appears to be inadequate to achieve this. A long
ash blanket residence time may instigate a stable change to
the ocean-atmosphere system (Pollack et al. 1976) and
amplify the initial forcing (Zielinski et al. 1996), but the
scale of disruption still seems less than required. Other
volcanogenic forcings such as changes in atmospheric
chemistry, soil respiration and the planktonic response to
both cooling and the addition of nutrients adsorbed onto
562 M. T. Jones et al.: The climatic impact of supervolcanic ash blankets
123
ash particles may be capable of initiating a long-term cli-
matic response, either solely or combined with other
feedbacks, but these have not been studied in detail. Cli-
mate variability may also dictate that long-term climate
change is only achievable at certain times, possibly during
susceptible periods of Milankovich cycles. It is also plau-
sible that the stresses of glacial buildup and rapid melting
may act as a cause, rather than an effect, to supervolcanism
(Zielinski et al. 1997). Explosive super-eruptions occur too
frequently to be solely responsible for mass extinction
events or to be able to completely destroy continental
ecosystems. They will, however, create a significant dis-
ruption to the climate in the short-term and possibly over
longer timescales. We hope to extend our studies to include
these effects.
Acknowledgments M.T.J. is funded by a NERC studentship.
R.S.J.S. is supported by a Royal Society Wolfson Award. Thanks to
the NERC Field Spectroscopy Facility at the University of Edinburgh
for use of equipment and the NERC Centre of Atmospheric Sciences
for high performance computer time. Thanks to Peter Baines, Vernon
Manville, Jessica Trofimovs, Matthew Watson, Fred Witham and
three anonymous referees for helpful comments.
References
Ambrose SH (1998) Late Pleistocene human population bottlenecks,
volcanic winter, and differentiation of modern humans. J Human
Evol 34:623–651
Baines PG, Sparks RSJ (2005) Dynamics of giant volcanic ash clouds
from supervolcanic eruptions. Geophys Res Lett 32:L24808
Bekki S (1995) Oxidation of volcanic SO2: a sink for stratospheric
OH and H2O. Geophys Res Lett 22(8):913–916
Bekki S, Pyle JA, Zhong W, Toumi R, Haigh JD, Pyle DM (1996)
The role of microphysical and chemical processes in prolonging
the climate forcing of the Toba eruption. Geophys Res Lett
23(19):2669–2672
Bishop J (2002) Early primary succession on Mount St Helens:
impact of insect herbivores on colonizing lupines. Ecology
83:191–202
Bush MB (2006) Ecology of a changing planet, 3rd edn. Prentice
Hall, New Jersey
Carroll MR (1997) Volcanic sulfur in the balance. Nature 389:543–
544
Collins M (2000) Understanding uncertainties in the response of
ENSO to greenhouse warming. Geophys Res Lett 27(21):3509–
3512
Collins BD, Dunne T (1986) Erosion of tephra from the 1980 eruption
of Mount St. Helens. Geol Soc Am Bull 97(7):896–905
Collins M, Tett SFB, Cooper C (2001) The internal climate variability
of HadCM3, a version of the Hadley Centre coupled model
without flux adjustments. Clim Dyn 17(1):61–81
Fagan W, Bishop J (2000) Trophic interactions during primary
succession: herbivores slow a plant reinvasion at Mount St.
Helens. Am Nat 155:238–251
Frogner P (2004) Al-F toxicity contributes to ecosystem stress and
species decline following large-scale volcanism. Geochim Cos-
mochim Acta 68(11):A147
Frogner P, Gıslason SR, Oskarsson N (2001) Fertilizing potential of
volcanic ash in ocean surface water. Geology 29(6):487–490
Gordon C, Cooper C, Senior CA, Banks H, Gregory HM, Johns TC,
Mitchell JFB, Wood RA (2000) The simulation of SST, sea ice
extents and ocean heat transports in a version of the Hadley
Centre coupled model without flux adjustments. Clim Dyn
16:147–168
Gregory JM, Ingram WJ, Palmer MA, Jones GS, Stott PA, Thorpe
RB, Lowe JA, Johns TC, Williams KD (2004) A new method for
diagnosing radiative forcing and climate sensitivity. Geophys
Res Lett 31:L03205
Hewitt SD, Broccoli AJ, Mitchell JFB, Stouffer RJ (2001) A coupled
model study of the last glacial maximum: was part of the North
Atlantic relatively warm? Geophys Res Lett 27:1571–1574
Johns TC, Gregory JM, Ingram WJ, Johnson CE, Jones A, Lowe JA,
Mitchell JFB, Roberts DL, Sexton DMH, Stevenson DS, Tett
SFB, Woodgate MJ (2003) Anthropogenic climate change for
1860 to 2100 simulated with the HadCM3 model under updated
emissions scenarios. Clim Dyn 20(6):583–612
Jones GS, Gregory JM, Stott PA, Tett SFB, Thorpe RB (2005) An
AOGCM simulation of the climatic response to a volcanic super-
eruption. Clim Dyn 25(7–8):725–738
Knight T, Chase J (2005) Ecological succession: out of the ash. Curr
Biol 15(22):R926–R927
Leavesley GH, Lusby GC, Lichty RW (1989) Infiltration and erosion
characteristics of selected tephra deposits from the 1980 eruption
of Mount St Helens. Hydrol Sci J (Journal Des Sciences
Hydrologiques) 34(3):339–353
Lee M-Y, Chen C-H, Wei K-Y, Iizuka Y, Carey S (2004) First Toba
supereruption revival. Geology 32(1):61–64
Luder T, Benz W, Stocker TF (2003) A model for long-term climatic
effects of impacts. J Geophys Res 108(10-1 to 10-17).
doi:10.1029/2002JE001894
Manville V (2002) Sedimentary and geomorphic responses to
ignimbrite emplacement: readjustment of the Waikato River
after the A.D. 181 Taupo Eruption, New Zealand. J Geol
110:519–541
Mason BG, Pyle DM, Oppenheimer C (2004) The size and frequency
of the largest explosive eruptions on Earth. Bull Volcanol
66:735–748
Nammah H, Larsen FE, McCool DM, Fritts R, Molnau M (1986) Mt.
St. Helens volcanic ash: effect of incorporated and unincorpo-
rated ash of two particle sizes on runoff and erosion. Agric
Ecosyst Environ 15:63–72
Oppenheimer C (2002) Limited global change due to the largest
known Quaternary eruption, Toba ~74kyr bp? Quat Sci Rev
21:1593–1609
Perkins ME, Nash BP (2002) Explosive silicic volcanism of the
Yellowstone hotspot: the ash fall tuff record. Geol Soc Am Bull
114(3):367–381
Pollack JB, Toon OB, Sagan C, Summers A, Baldwin B, Van Camp
W (1976) Volcanic explosions and climate change: a theoretical
assessment. J Geophys Res 81:1071–1083
Pope VD, Gallani ML, Rowntree PR, Stratton RA (2000) The impact
of new physical parameterizations in the Hadley Centre climate
model—HadAM3. Clim Dyn 16:123–146
Pyle DM (1995) Mass and energy budgets of explosive volcanic
eruptions. Geophys Res Lett 5:563–566
Pyle DM (2000) Sizes of volcanic eruptions. In: Sigurdsson H,
Houghton BF, McNutt SR, Rymer H, Stix J (eds) Encyclopedia
of volcanoes. Academic, New York, pp 263–269
Rampino M (2002) Supereruptions as a threat to civilizations on
earth-like planets. Icarus 156:562–569
Rampino MR, Ambrose SH (2000) Volcanic winter in the Garden of
Eden: the Toba supereruption and the Late Pleistocene human
population crash. Geol Soc Am Spec Paper 345:71–82
Rampino M, Self S (1992) Volcanic winter and accelerated glaciation
following the Toba super-eruption. Nature 359:50–52
M. T. Jones et al.: The climatic impact of supervolcanic ash blankets 563
123
Rampino M, Self S (1993) Climate-volcanism feedback and the Toba
Eruption of ~74000 years ago. Quat Res 40:269–280
Reynolds RL (1975) Comparison of TRM of Yellowstone group and
DRM of equivalent air-fall ash beds in western United States.
Trans Am Geophys Union 56(12):977–977
Robock A (2000) Volcanic eruptions and climate. Rev Geophys
38(2):191–219
Rose WI, Chesner CA (1987) Dispersal of ash in the great Toba
eruption, 75 ka. Geology 15:913–917
Savarino J, Bekki S, Cole-Dai J, Thiemens MH (2003) Evidence from
sulfate mass independent oxygen isotopic compositions of
dramatic changes in atmospheric oxidation following massive
volcanic eruptions. J Geophys Res Atmos 108(D21):4671
Sparks RSJ, Bonnecaze R (1993) Sediment-laden gravity currents with
reversing buoyancy. Earth Planet Sci Lett 114(2–3):243–257
Sparks RSJ, Walker GPL (1977) The significance of crystal-enriched
air-fall ashes associated with crystal-enriched ignimbrites. J
Volcanol Geotherm Res 2:329–341
Sparks RSJ, Moore JG, Rice CJ (1986) The initial giant umbrella
cloud of the May 18, 1980 explosive eruption of Mount St.
Helens. J Volcanol Geotherm Res 28:257–274
Sparks RSJ, Bursik MI, Carey SN, Gilbert JS, Glaze LS, Sigurdsson
H, Woods AW (1997) Volcanic Plumes. Wiley, New York
Sparks RSJ, Self S, Grattan J, Oppenheimer C, Pyle DM, Rymer H
(2005) Supereruptions: global effects and future threats. Geo-
logical Society of London
Stott PA, Tett SFB, Jones GS, Allen MR, Mitchell JFB, Jenkins GJ
(2000) External control of 20th century temperature by natural
and anthropogenic forcing. Science 290:2133–2137
Tett SFB, Jones GS, Stott PA, Hill DC, Mitchell JFB, Allen MR,
Ingram WJ, Johns TC, Johnson CE, Jones A, Roberts DL, Sexton
DMH, Woodgate MJ (2002) Estimation of natural and anthro-
pogenic contributions to 20th century climate change. J Geophys
Res 107:ACL 10-1 to ACL 10-24. doi:10.1029/2000JD000028
Timmreck C, Graf H-F (2005) The initial dispersal and radiative
forcing of a Northern Hemisphere mid latitude super volcano: a
Yellowstone case study. Atmos Chem Phys Discuss 5:7283–
7308
Trenberth K (1999) Atmospheric moisture recycling: role of advec-
tion and local evaporation. J Clim 12:1368–1381
Wilson MF, Henderson-Sellers A (1985) A global archive of land
cover and soils data for use in general circulation climate
models. J Climatol 5(2):119–143
Witham C, Oppenheimer C, Horwell CJ (2005) Volcanic ash-
leachates: a review and recommendations for sampling methods.
J Volcanol Geotherm Res 141(3–4):299–326
Woods AW, Wohletz K (1991) Dimensions and dynamics of co-
ignimbrite eruption columns. Nature 350:225–228
Zielinski GA, Mayewski PA, Meeker LD, Whitlow S, Twickler MS,
Taylor K (1996) Potential atmospheric impact of the Toba mega-
eruption ~71,000 years ago. Geophys Res Lett 23(8):837–840
Zielinski GA, Mayewski PA, Meeker LD, Gronvold K, Germani MS,
Whitlow S, Twickler M, Taylor K (1997) Volcanic aerosol
records and tephrochronology of the Summit, Greenland, ice
cores. J Geophys Res 102(C12):26625–26640
564 M. T. Jones et al.: The climatic impact of supervolcanic ash blankets
123
